The angiomotins – from discovery to function

Abstract

Angiomotins were originally identified as angiostatin binding proteins and implicated in the regulation of endothelial cell migration. Recent studies have shed light on the role of Angiomotins and other members of the Motin protein family in epithelial cells and in pathways directly linked to the pathogenesis of cancer. In particular, Motins have been shown to play a role in signaling pathways regulated by small G-proteins and the Hippo-YAP pathway. In this review the role of the Motin protein family in these signaling pathways will be described and open questions will be discussed.

The Angiomotins – Discovery

Sequencing of cDNA clones from human brain cDNA libraries identified a 5061 base pair (bp) cDNA clone, designated KIAA1071, with an open reading frame (ORF) coding for a 473 amino acid protein [1]. However, it was not until 2001 through a yeast-two hybrid screen of a human placenta cDNA library, using the kringle domains 1 to 4 of angiostatin as bait that Angiomotin, the founding member of the motin family, was first cloned [2]. Northern blot analysis detected two transcripts at 6.5 kb and 7.5 kb in a broad spectrum of analyzed tissues. Given the predominant expression of Angiomotin in endothelial cells and its involvement in mediating the anti-migratory properties of angiostatin; it was given its name [2]. With a cytogenetic location at chromosome Xq23, it shares 85% homology with the mouse Angiomotin ortholog, and was given the HUGO nomenclature gene designation AMOT. Angiomotin is a 675 amino acid protein and with an estimated molecular mass of 80 kDa, it was termed Amot-p80 [2, 3]. Subsequent analysis of GenBank databases identified 2 additional polypeptides that show significant sequence homology to Amot. These were Angiomotin-like 1 (AmotL1, a.k.a. JEAP) and Angiomotin-like 2 (AmotL2, a.k.a. MASCOT). JEAP (junction-enriched and –associated protein) was initially identified in a screen for novel tight-junction (TJ)-associated proteins by a fluorescence localization-based expression cloning method [4] and due to similarity to Angiomotin, subsequently named Angiomotin-like 1 [5]. Human AmotL1 (hAmotL1) (NCBI Accession NM_130847) is a 956 amino acid protein with a predicted molecular mass of approximately 106 kDa. The human AmotL1 (hAmotL1) gene is located at chromosome 11q21 and shares 85% sequence homology with mouse AmotL1 (mAmotL1), an 882 amino acid protein (isoform 2, Uniprot database). Angiomotin-like 2 (AmotL2) (NCBI Accession NM_016201) is a 780 amino acid protein with a predicted molecular mass of 86 kDa. The gene is mapped to chromosome 3q21-q22 [5], and like AmotL1 the coding sequence shares 85% homology with mouse AmotL2, a 772 amino acid protein. Functional characterization of AmotL2 showed it colocalizes with MAGI-1 at epithelial tight junctions, an interaction found to be dependent on the AmotL2 coiled-coil domain. The protein was therefore named MAGI-1-associated coiled-coil tight junction protein (MASCOT) [6]. In the interest of simplicity we will subsequently refer to these as AmotL1 and AmotL2. Since these proteins exhibit such significant sequence homology, suggestive of a common evolutionary origin, this family of proteins was named Motins [5]. Later studies identified an additional form of Amot called p130 (Amot-p130), which arises from alternative splicing of the Amot gene between exons 2 and 3 and carries an extended 409 amino acid N-terminus [3]. Amot-p130 (NCBI Accession NM_001113490) is a protein composed of 1084 amino acids with an estimated molecular mass of 130 kDa. The apparent lack of proteolytic cleavage sites suggests that Amot-p80 is not generated from Amot-p130 via hydrolytic catalysis, but that the Amot gene produces both Amot isoforms through alternative splicing [3].

The Motin protein family members share several structural characteristics (Figure 1). The N-terminus, shared between Amot-p130, AmotL1, and AmotL2, is composed of conserved glutamine-rich domains, PPxY motifs (Amot: ²³⁹PPEY²⁴² and ²⁸⁴PPEY²⁸⁷; AmotL1: ³¹⁰PPEY³¹³ and ³⁶⁷PPEY³⁷⁰; AmotL2: ²¹⁰PPQY²¹³ and the slightly divergent ²⁵²PPVF²⁵⁵), plus a recently identified unconventional LPTY motif lying upstream of the other two (Amot: ¹⁰⁶LPTY¹⁰⁹; AmotL1: ¹⁸⁸LPTY¹⁹¹; AmotL2: ¹⁰⁴LPTY¹⁰⁷) [7, 8]. Of note, AmotL2 differs from the other members at one of the PPxY motifs, as the tyrosine residue is replaced by phenylalanine. Remarkably these motifs are highly conserved, not only among the Motin family members, but also across species [8]. Since Amot-p80 lacks the entire N-terminal, these motifs are not present. The conserved coiled-coil (CC) domain and the C-terminal PDZ-binding domain compose the C-terminal region. The predicted locations for the coiled-coil domains are as follows: Amot-p130 (429 aa – 689 aa; 721 aa – 751 aa); AmotL1 (438 aa – 639 aa; 665 aa – 694 aa; 729 aa – 762 aa); AmotL2 (308 aa – 581 aa) (Uniprot database). Importantly, the N-terminal 242 amino acids of Amot-p80 was suggested to encode for a positively charged CC fold due to its positional conservation with amphiphysin (bin/amphiphysin/rvs) BAR domain [9, 10]. Since this domain is a conserved region across the Motin family members it was termed Amot coiled-coil homology (ACCH) domain [11]. The coiled-coil regions comprise two or more right-handed α-helices wrapped around each other with a left-handed superhelical twist [12]. CC domains contribute to several biological and structural functions, including oligomerization. Oligomeric regulation has been described for all of the members of the Motin family, either by forming homo-oligomers through self-association, or hetero-oligomers with other family members through their CC domains [6, 13, 14]. Between the conserved CC domain and the C-terminal PDZ-binding region is localized an angiostatin-binding hydrophobic domain, present in Amot-p80 and Amot-p130, yet absent in AmotL1 and AmotL2 [2, 5]. The consensus motif for the PDZ domain binding is highly conserved, and its presence in the Motins’ structure offered the first clue for their potential involvement in signaling pathways.

Fig. 1. The Motin protein family.

Angiomotin p80 (Amot-p80) is an N-terminal truncated version of Angiomotin p130 (Amot-p130) as a result of alternative splicing. Angiomotin-like proteins 1 and 2 (AmotL1 and AmotL2), share sequence identity to Amot-p130, but notably lack the angiostatin-binding domain. * depicts phosphorylation sites with available experimental confirmation. ** depicts ubiquitination sites with available experimental confirmation. CC/BAR – Coiled-Coil/(Bin/Amphiphysin/Rvs) domain. PDZ – Post synaptic density protein (PSD95), Drosophila disc large (Dlg1) and Zonula occludens-1 (ZO-1) domain.

In terms of protein topology, a number of models have been proposed [3, 15, 16]. Based on studies in mouse aortic endothelial (MAE) cells, in which an antibody used against the angiostatin-binding domain was effective without any prior membrane permeabilization step, it was proposed that Angiomotin localizes to the cell membrane as a transmembrane protein with both isoforms forming a transmembrane loop, the angiostatin-binding domain oriented outwards and the N- and C-terminal domains in an intracellular orientation [16]. DNA vaccination of mice, targeting human Amot-p80, generated antibodies that bind to Amot-p80 on the endothelial cell surface. These studies further support a topology where the angiostatin-binding domain is in an extracellular orientation [17, 18]. In contrast, in studies with bile canaliculi fractions prepared from mouse liver without the use of detergent, endogenous Amot-p130, Amot-p80, and AmotL2 were extracted with intracellular fractions, suggesting that these Motins are not trans-membrane proteins [2, 15]. Interestingly, recent studies demonstrated the exclusive binding of angiostatin to glioblastoma cells that express Angiomotin. While both proteins appear to be present on the cell surface they do not appear to colocalize, suggesting Angiomotin mediates angiostatin binding indirectly.

The structural similarities between the Motins and the absence of an obvious angiostatin-binding domain in AmotL1 and AmotL2 might suggest that this family of proteins interacts with the hydrophobic phospholipid membrane bilayer indirectly. Moreover, protein topology prediction algorithms, such as Topcons [19] also paint a complex picture predicting the presence of potential hydrophobic motifs featuring a transmembrane topology but no predicted extracellular domains. Thus, further experimental analysis is needed to clarify these questions.

Although there is a high degree of similarity between the Motin proteins, there are functional differences that are still not fully understood. One clue to this is the finding that the Motins are differently expressed across different tissues and, if expressed in the same tissue, exhibit variable spatiotemporal and expression level patterns. Analysis of mRNA expression of Amot, AmotL1, and AmotL2 in diverse human tissues, based on data generated by the Illumina human BodyMap 2.0 platform, demonstrates that all family members are expressed in the surveyed tissues, yet at variable expression levels (Figure 2). Amot showed higher transcript levels in the testis, followed by the brain and thyroid. The lowest expression is seen in the liver and adrenal gland. AmotL1 is highly expressed in skeletal muscle, representing the highest mRNA expression level across the entire analysis. The lowest levels are found in the blood. AmotL2 expression is highest in breast and lowest in the liver. This analysis is in agreement with previous studies examining Motin tissue expression patterns not only in human but also in mouse. Specifically, Amot expression was found in mouse and rat skeletal muscle [5, 20], highly expressed in mouse brain, pancreas, and salivary gland [3, 5, 15]. Similar to its human homologue, AmotL1 mRNA was predominantly expressed in mouse skeletal muscle [5], pancreas, and salivary gland [15]. The expression pattern of the Motins is therefore tissue-type dependent, predicting functional variability within the family.

Fig. 2. RNA expression profiles of Motin family members in human tissue.

Results are based on data generated by the Illumina human BodyMap 2.0 platform. RNA obtained from 16 normal human tissues was sequenced and mapped. Fragments Per Kilobase of exon per Million fragments mapped (FPKM) were calculated using the Cufflinks program and expression levels scaled using (100 × FKPM)^1/2 (http://cufflinks.cbcb.umd.edu/). Major tissues are colored according to 6 categories: Immune (red), Nervous (green), Muscle (yellow), Internal (blue), Secretory (violet), and Reproductive (turquoise).

The expression patterns of Motin family members are also variable during development. Amot, AmotL1, and AmotL2 mRNAs showed spatiotemporal-dependent expression in the mouse uterus during pre-implantation and post-implantation periods [21-23]. At a very early stage of pregnancy, Amot and AmotL1 transcripts were initially expressed in the luminal epithelium and later on in the stroma. Alteration of their expression pattern occurred within a fairly small timeframe, specifically at 4 days post-pregnancy. The same expression shift was observed during the post-implantation period. Intriguingly, AmotL2 expression was barely detected in either period except on day 1 of pregnancy [21]. AmotL2 might therefore represent an evolutionary outlier of the Motin family, a concept initially proposed following phylogenetic analysis [5].

Finally, while several cell lines have been used in the functional studies of the Motin family, in many of these the expression of endogenous Motin family members could not be detected using currently available antibodies and/or qPCR. In some cases, studies from different groups report contradictory findings in the same cell line, suggesting experimental conditions could significantly impact expression of Motin proteins. Thus, caution should be exercised in choosing the appropriate cell type and experimental conditions for analysis. Table 1 summarizes expression of Motin family members in cell lines based on the published literature.

Table 1.

Expression levels of Motin family members in different cell lines.

Cell lines	Amot-p130	AmotL1	AmotL2	Analytic Methodology	Ref.
Epithelial
ACHN	-	-	Unknown	WB	[47]
BT474	+/−	-	Unknown	WB	[46,48]
BT549	-	-	Unknown	WB	[48]
HCC1937	+	Unknown	Unknown	WB	[46]
HEK293	+	+/−	+/−	WB/qPCR	[47,48]
HEK293T	+	-	+/−	WB/qPCR	[46,47]
HeLa	-	-	+	WB/qPCR	[47]
HEpG2	-	-	Unknown	WB	[48]
Hs-578T	+/−	-	Unknown	WB	[46,48]
H2.35	-	+	Unknown	WB	[47]
MCF7	+/−	-	Unknown	WB	[46,48]
MCF10A	-	-	+	WB/qPCR	[46,47,48]
MDA-MB-231	+/−	-	Unknown	WB	[46,48]
MDA-MB-436	+	Unknown	Unknown	WB	[46]
MDA-MB-468	+	Unknown	Unknown	WB	[46]
MDCK	+/−	-	+	WB/qPCR	[9,46,47]
SK-Br3	+/−	-	Unknown	WB	[46,48]
T47D	+/−	-	Unknown	WB	[46,48]
ZR75	+	Unknown	Unknown	WB	[46]
Endothelial
BCE	+	-	Unknown	WB	[13]
HUVEC	-	+	Unknown	WB	[13]
MAE	-	+	Unknown	WB	[13]
MS-1	+	+	Unknown	WB	[13]
Pmt-EC	+	+	Unknown	WB	[13]
TEC	+	+	Unknown	WB	[13]
Fibroblast/Fibroblast-like
Cos-7	-	-	+	WB/qPCR	[46]
MEF	+	+	Unknown	WB	[46]
NIH-3T3	-	+	+	WB/qPCR	[46]
Neural
SC4	+	Unknown	Unknown	WB	[7]
RT4	+	Unknown	Unknown	WB	[7]
R3	+	Unknown	Unknown	WB	[7]
MPNST 90-8TL	+	Unknown	Unknown	WB	[7]

The endogenous expression of the Motins is based on studies that specifically screened for Motin expression across various cell lines. Motins are showed solely as present or absent in the respective cell line regardless of their expression levels. (-) Absent. (+) Present. (+/−) Inconsistent data between detection approaches. Unknown indicates that the expression of the Motin family member has not been tested in that particular cell line. WB – Western blotting. qPCR – quantitative PCR.

Regulation of the Angiomotins

A key post-transcriptional regulatory mechanism of the Motins appears to be by alternative splicing. The first evidence for the occurrence of coding sequencing diversity surfaced with the description of two splice isoforms at the hAmotL1 locus, the first lacking exon 2 and the second exons 2 and 3 [24]. It was predicted that only the first isoform coded for a functional protein of 906 amino acids with a predicted molecular weight of 90 kDa [24]. Uniprot protein sequence database predicts two alternative transcripts from the hAmotL1 locus, one with 956 amino acids coding for a 106 kDa protein, and a second one generated by a 50 amino acid deletion on AmotL1 N-terminal with a molecular mass around 101 kDa. Regarding Amot, several studies have broadly characterized the two known splice isoforms, Amot-p80 and Amotp130. As previously stated, Amot-p130 carries an extended N-terminal of 409 amino acids and this is the only structural difference from Amot-p80 [3]. The N-terminal region alone is sufficient to generate temporal, spatial, and functional differences between isoforms [3, 9, 14, 20, 25]. Even though Amot has only two transcript variants described so far, additional bands corresponding to proteins of 70 and 110 kDa were observed in western blotting experiments by different groups [14, 20, 26]. Whether these bands represent functional or a transient unknown Amot isoforms remains to be determined. To the best of our knowledge, besides the 780 amino acid protein [5], the only alternative splicing event described for the hAmotL2 locus is a 466 amino acid form lacking the 313 amino acids located on the N-terminal of AmotL2 [24, 27]. Although the molecular mass of this short isoform is not mentioned in the study [27], commercial antibodies developed against AmotL2 detect a 51kDa protein that potentially represents this isoform (Sigma-Aldrich; GenWay Biotech; AvivaSystemsBiology; and others). Uniprot further predicts three additional human isoforms, two of which have no available experimental confirmation. Specifically, a 779 amino acid peptide with predicted molecular mass of 85 kDa; a 777 amino acid protein generated by a two residue deletion at position 525-526 (encodes a protein with predicted molecular mass of 85 kDa); and finally a 837 amino acid protein generated by a 58 amino acid insertion, starting at the N-terminus, predicted to encode a 92 kDa protein. In mice, multiple transcript variants of the Motin family are predicted and we refer to the available databases for further analysis.

Other mechanisms involved in the regulation of the Angiomotins have only recently begun to come to light. Previous omics-type approaches identified a multitude of post-translational modifications of the Motin proteins, including phosphorylation, ubiquitination, acetylation, and glycosylation. In some cases, the Motins were identified as substrates for specific enzymes, such as the ataxia telangiectasia mutase (ATM), ataxia telangiectasia and Rad3-related (ATR) or cyclin-dependent kinase 2 (CDK2) kinases [28, 29]. However, these findings remain to be confirmed and have not been studied in depth. There are several phosphorylation sites predicted by the PhosphoSite database within the Motin family. Multiple hits were suggested and, for simplification, we will only discuss the sites confirmed experimentally. Lists of potential phosphorylation and ubiquitination sites (predicted by UbPred database) in human Motins are detailed in Table 2.

Table 2.

Potential Phosphorylation and Ubiquitination sites in human Motin family members.

	Amot-p130		AmotL1		AmotL2
Phosphorylation Sites	Tyr109	Ser787	Tyr191	Ser724	Thr10	Tyr573
	Ser175	Tyr836	Tyr218	Ser793	Tyr92	Thr596
	Ser305	Tyr842	Tyr219	Ser796	Lys100	Ser608
	Ser312	Tyr847	Ser225	Thr803	Tyr107	Ser609
	Tyr420	Thr848	Ser241	Ser805	Ser159	Ser661
	Tyr527	Ser852	Ser262	Ser809	Ser183	Ser753
	Tyr599	Ser855	Ser269	Ser828	Ser236	Ser759
	Ser712	Ser1059	Ser295	Ser900	Lys408	Ser762
	Ser714	Thr1061	Thr373	Thr901	Ser540
	Ser718		Ser374	Ser906
			Tyr661	Ser924
			Ser720
Ubiquitination Sites	Lys94	Lys520	Lys174	Lys326	Lys100
	Lys156	Lys525	Lys195	Lys649	Lys111
	Lys209	Lys535	Lys246	Lys881	Lys436
	Lys219	Lys545	Lys254	Lys929	Lys721
	Lys231	Lys553	Lys302	Lys943
	Lys255	Lys619
	Lys481

The phosphorylation and ubiquitination sites displayed are based on predictions made by Phosphosite and UbPred databases.

The post-translational modifications that have been studied in more detail are the ubiquitination at Lysine 481 and phosphorylation of Angiomotin p130 at serine 175. In HEK293T cells, Amot-p130 is ubiquitinated at residue Lys481 by Atrophin-1 interacting protein 4 (AIP4)/Itch, a member of the Nedd4 (neural-precursors-cell-expressed developmentally down-regulated)-like ubiquitin E3 ligases family. AIP4 ligase activity is triggered by direct interaction between AIP4 WW1 and WW2 domains and Amot-p130 PPxY motifs, which promotes either self or Amot-p130 ubiquitination [8, 30]. This regulation is mediated by the large tumor suppressor (LATS)1/2, a core component of the Hippo signaling pathway that also binds AIP4 [30]. Amot-p130 is thus a physiological substrate for the E3 ligase, a role extended to AmotL1 but not to AmotL2 or Amot-p80 [30]. Other members of the E3 ubiquitin ligase family, Nedd4 and Nedd4.2, also induced Amot-p130 and AmotL1 polyubiquitination and proteasome-dependent degradation. However, phosphorylation of Nedd4.2 by the non-receptor tyrosine kinase c-Abl was shown to inhibit AmotL1 degradation [31]. Recruitment of c-Abl is promoted by yes-associate protein (YAP), a transcriptional co-factor involved in Hpo-YAP signaling. It was found that YAP promotes Nedd4.2 phosphorylation by c-Abl, thus ensuring Nedd4.2 does not target AmotL1 for 26S proteasomal degradation [31]. Furthermore, through its WW domains, Nedd4 binds to Amot-p130 first and second L/PPXY motifs, akin to the YAP-Amot-p130 association. This suggests that YAP might compete with Nedd4 for Amot-p130 binding [8]. Whether YAP, by binding to Amot-p130, prevents its Nedd4 ubiquitin-dependent degradation is unknown and further studies are needed to address this question. Importantly, both direct (BT474, HEK293T, and MCF7 cell lines) and indirect (in MDA-MB-468 cells) correlations between AIP4 activity and Amot-p130 steady state levels were found to be cell-context dependent [8, 30].

Experimental confirmation of Motin family member phosphorylation was recently reported by multiple groups who identified LATS1/2-dependent phosphorylation of Amot-p130(Ser175), AmotL1(Ser262), and AmotL2(Ser159) at conserved HVRSLS motifs [23, 30, 32, 33]. LATS2 binds AmotL2 on its N-terminal 307 amino acids [34], and Amot-p130. The direct binding between LATS2 and Amot was enhanced by Amot Ser175 phosphorylation and is dependent on LATS kinase activity [23, 33]. The phosphorylation of Motins by LATS has been shown to occur both in vitro and in vivo [23, 30, 32, 33]. Finally, a model was proposed where both phosphorylation of Amot-p130 at serine 175 and ubiquitination at Lysine 481 act together as a major regulatory mechanism to mediate YAP activity and Hpo-YAP pathway signaling [30]. Under serum starvation conditions, activation of the Hippo pathway triggers Amot p130 phosphorylation by LATS1/2 at Ser175. AIP4 binding to phosphorylated Amot-p130 impedes the AIP4-LATS1/2 interaction and subsequent LATS degradation. Once associated, Amot-p130 is ubiquitinated by AIP4 at Lys481 increasing its steady state levels. The increased levels of LATS and Amot-p130 promote YAP ubiquitination [30]. The regulation of YAP by Amot-p130 is still not fully elucidated and will be discussed further below.

Functional redundancy in the Motin family

The extent of functional and genetic redundancy between the members of the Motin family is unknown. As discussed above, Motin expression patterns suggest both distinct and overlapping roles, however limited functional evidence exists so far. In the mouse embryo (zygote), simultaneous silencing of Amot/AmotL1/AmotL2 resulted in a remarkable increase in cells expressing the differentiation gene Cdx2⁺ compared to depletion of Amot alone. Thus genetic redundancy between the Motin family appears to occur in promotion of embryonic inner cell mass (ICM) differentiation [22]. Furthermore, simultaneous knockdown of the Motin proteins in HEK293T cells caused a striking up-regulation in the activity of a Wnt reporter when compared to knocking down each member alone, which resulted in the down-regulation of the Wnt signaling [27]. Together with other studies [8], these findings present initial evidence for the occurrence of both genetic and functional redundancy within the Motin family.

Function: Role in endothelial cell biology and angiogenesis

Given that the angiomotins were originally discovered as mediators of endothelial cell migration, early studies into the role of the Motin family during development focused on endothelial cells (EC) biology and blood vessel formation. Given the focus of this review, we will only cover these studies briefly.

Studies to examine the role of Motin family members in vivo have suggested these proteins have critical functions during development. Amot knockout 129/SvEv background mice displayed early lethality and most mice died soon after gastrulation due to defects in cell migration [35]. The severity of this phenotype is dependent on genetic background, as crossing the Amot^+/− 129/SvEv mice to C57/B6 to generate a mixed 129/B6 background resulted in a delayed lethality of the Amot^−/− mice (75% die between E11 and E15). Careful analysis of these mice indicated that Amot plays a critical role in blood vessel formation through regulation of endothelial cell migration and polarity. Similarly, the knockout of Amot in Zebrafish embryo also impaired endothelial cell migration and blood vessel formation [36]. Efforts to manipulate the expression of other members of the Motin family have also been reported. The knockdown of AmotL1 in Zebrafish embryos yields a phenotype similar to Amot knockdown with a migratory defect of endothelial cells and vessel formation, although the mechanisms might differ [13]. Concerning AmotL2, knockdown in Zebrafish embryos resulted in cell migration and proliferation defects [37, 38]. Overall, the effect of knocking down either one or several members of the Motin family leads to defective phenotypes such as impaired EC migration, polarization, and proliferation.

The regulation of endothelial cell migration and proliferation by Angiomotins requires their C-terminal PDZ-binding domains. Deletion of those amino acids in AmotL2/Amot-p80 knockdown Zebrafish embryos was sufficient to restrict migration [38, 39]. In mouse embryos, the deletion of Amot PDZ-binding domain under the EC-specific Tie promoter inhibited the response of MAE cells to chemotactic cues resulting in lethality by day E9.5 [25]. Mechanistic studies show that Amot, by binding the polarity complex Patj/Mupp1:Pals1, indirectly associates with Syx, a RhoA GTPase exchange factor (RhoGEF) binding-domain protein, forming a ternary complex [39, 40]. Specifically, Amot directly binds to the PDZ3 domain of Patj/Mupp1, whilst Syx mostly interacts with Patj/Mupp1 PDZ10 domain. AmotL1 and AmotL2 also associate with Patj/Mupp1 via its PDZ1 and PDZ2 domains, respectively [15, 39]. Importantly, Amot was shown to strongly associate with monophosphorylated phosphatidylinositols PI(4)P and PI(3)P [11]. Thus, it is suggested that Amot, by interacting through its PDZ-binding domain with Patj/Mupp1:Syx polarity complex, mediates its trafficking through endocytic vesicles from EC junctions to the leading edge of endothelial migrating EC resulting in recruitment of Syx activity to the leading edge. The targeting of the endocytic vesicles is likely mediated by Amot ACCH domain (lipid-binding domain) [39, 41].

Function: Role in small G-protein signaling

Since Motins have pivotal functions during development, EC migration and proliferation, additional studies probed whether epithelial cell polarity is regulated by Amot. Interestingly, all members of the Motin family have been shown to associate with tight junctions through binding to the tight junction-associated proteins Patj and Pals1 [9, 15, 39]. Patj and Pals1 are cytoplasmic scaffolding proteins that together with the transmembrane protein CRB3 form the apical CRB complex [42]. Work conducted by us and by others showed that in mammalian epithelial cells, Amot interacts with Rich1, a Cdc42/Rac1 GTPase Activating Protein (GAP), via their mutual BAR/CC domains [9, 43]. This association targets Amot to the TJs, wherein by binding Patj, Amot forms a complex composed by Amot:Patj:Pals1:Rich1 localized at the apical membrane. Importantly, Amot negatively regulates Rich1 GAP activity, which has been suggested to impair TJ integrity [9]. In addition, Amot directly binds to Merlin, a negative mediator of Rac1-signaling, also through their mutual BAR/CC domains [43]. Thus a model emerges in which Amot plays a role in mediating Rac1 signaling at junctional structures as Merlin and Rich1 compete for Amot binding. Merlin, by releasing Rich1 from the Amot inhibitory complex, facilitates Rich1 GAP activity of Rac1 and subsequent signal transduction through direct downstream effectors such as the p21-activated kinases and MAPK pathway (Figure 3) [43]. This coordination was also observed in the pathological setting of neurofibromatosis type 2 (NF2), a condition caused by the loss of the NF2 gene, which codes the Merlin protein [44, 45]. Amot knockdown in Nf2^−/− Schwann cells inhibits proliferation in vitro and leads to impaired tumor formation in vivo [43]. Other studies have also reported positive regulation of the ERK/MAPK pathway by Amot and AmotL2 in different cell contexts such as Zebrafish embryos, angiogenic endothelial cells (HUVEC) [37], or epithelial breast cancer cell lines like MCF7, SKBR3 or MDA-MB-468 [46]. Conversely, knockdown of AmotL2, but not AmotL1, in MCF10A cells activated both AKT and ERK/MAPK pathways, suggestive of a negative, rather than positive, mediation of these pro-proliferation pathways [37]. The differences in consequences of AmotL2 knockdown could be attributed to cell-type specificity. As AmotL2 mRNA and protein levels are barely detectable in MCF10A cells [47, 48] (Table 1), further studies are needed to gain a clear understanding of AmotL2 function in these cells.

Fig. 3. Angiomotin, Rich1, and Merlin regulate Rac1/Cdc42 activity.

A model showing mutually-exclusive interactions between Angiomotin with Merlin or Rich1 that modulate Rac1/Cdc42 activity. Merlin, by binding Angiomotin through their mutual CC domains, releases Rich1 from the inhibitory Amot-Rich1 complex. Rich1 is then able to inactivate Rac1/Cdc42 by converting Rac1/Cdc42-GTP to Rac1/Cdc42-GDP.

Function: Angiomotin regulation of Hpo/YAP signaling

The Hpo/YAP pathway has been described in a number of recent reviews [49-51]. Briefly, this pathway was initially characterized in flies and is highly conserved in mammalian cells. It plays a central role in various cellular behaviors including proliferation, survival, and cell contact inhibition[52-54]. The pathway is composed of a core kinase cascade, in which the Mst1/2 kinases (Hippo in flies), in complex with scaffold protein WW45 (Salvador in flies), phosphorylate LATS1/2 (Warts in flies) and Mob1 (Mats in flies). Phosphorylated LATS1/2 binds to and further activates the Mob1 kinase, which in turn, phosphorylates YAP (Yorkie in flies), a transcriptional co-activator. The phosphorylation of YAP prevents it from entering into the nucleus, where it can form transcriptionally active complexes with TEAD (Scalloped in flies) and other transcription factors to drive the expression of proproliferative or anti-apoptotic genes. A series of studies have demonstrated that akin to what has been observed in flies, the mammalian Hpo-YAP pathway can also regulate organ size. For example, inducible overexpression of YAP in adult mouse liver results in rapid increase in liver size [55, 56]. Comparable hepatomegaly phenotypes were observed when Mst1, Mst2, and WW45 were ablated specifically in the liver [57-59]. Finally, increased YAP activity appears to be a common occurrence in human hepatocellular carcinoma [36].

Recently there has been a flurry of reports implicating Motin family members in the regulation of the Hpo/YAP signaling in mammalian cells. The Drosophila Hippo signaling cascade is highly conserved throughout evolution, and functional conservation was confirmed in many cases by the ability of mammalian orthologs to rescue the corresponding Drosophila mutants in vivo [60, 61], although this is not observed in all cases [62]. Remarkably, the Motin family members do not appear to have Drosophila orthologs [63]. The protein Expanded (Ex), which shares significant sequence homology to both Amot-p130 and to FRMD6/Willin, was proposed to represent a functional equivalent in Drosophila [62, 64, 65]. Ex PPXY motifs directly interact with Yorkie (Yki) WW domains forming a complex localized at the apical junctions of cells in the eye imaginal disc [66-68]. This mechanism of interaction, as discussed below, resembles the physical association between Amot and AmotL1 with YAP and TAZ. However, FRMD6/Willin alone was not able to rescue Ex loss in Drosophila [62], thus raising the possibility that Ex function has been split during evolution into the cooperative actions of Angiomotins and FRMD6/Willin.

Elegant work conducted by Sowa et al., (2009) first described Amot as a YAP binding partner. This finding was part of a global proteomic survey analyzing deubiquitinating enzymes and their interacting protein complexes in a mammalian cell system [69]. Several studies have subsequently demonstrated that Amot-p130 and AmotL1 directly interact with YAP via Motins LPTY and first PPEY motif (Amotp130: ¹⁰⁶LPTY^{109 239}PPEY²⁴²; AmotL1: ¹⁸⁸LPTY^{191 310}PPEY³¹³), and mostly the first WW domain of YAP (localized between 171 to 208 residues of YAP [70]. This binding is independent of their second PPEY motif and PDZ-binding domain as mutations in these regions showed only mild effects [7, 31, 37, 47, 48, 71]. Yet considerable decrease in the Amot-YAP interaction was observed with Amot coiled-coil (CC) domain mutants [47]. Regarding AmotL2, its PPXY motif (²¹⁰PPEY²¹³) was shown to bind YAP mainly through the first WW domain, and independently of the PDZ-binding motif [34, 37]. Noteworthy, the association of Amot-p130, AmotL1, and AmotL2 to TAZ (YAP1 paralog) occurs exclusively via the Motin PPXY motif and TAZ WW domain [48, 72]. Consistently, neither YAP nor TAZ show interaction with the short isoform Amot-p80. Therefore, at least under the conditions used in the above-mentioned analyses, the Motin family members represent a strong binding partner of YAP. Interestingly, several other proteins including LATS1/2 also bind to the WW domains of YAP through their PPXY motifs [73-78].

It should be noted that some potential discrepancies exist in the literature as to which isoform of YAP interacts with the Motins. Some studies have shown that the Motin family members bind YAP isoform 2, which has two WW domains, but not to isoform 1, with one WW domain [30, 71, 79]. The difference between the two isoforms consists of an extra 38 amino acids that compose the second WW domain. Intriguingly, a number of other studies have identified an interaction between Motins and the YAP1 isoform. These differences could result from multiple reasons such as different experimental systems or protein expression beyond physiological levels. In any case, caution is advised when determining which YAP isoform is being used in a given experimental setup. Indeed, at least two out of the eight YAP isoforms appear to differently regulate the Hpo/YAP signaling pathway [71, 74].

Besides YAP, the Motins also associate with other effectors of the Hpo/YAP signaling pathway. Specifically, Merlin interacts with Amot-p130 through their mutual CC domains [43]; LATS2 associates with Amot-p130 CC domain, and with the N-terminal 307 amino acids of AmotL2 through its kinase domain, including the adjacent Mob-1 binding region [23, 34]. AmotL2 also binds MST2, yet the domains involved in this interaction are currently unknown [34]. KIBRA, an upstream component of the pathway, binds to the second ²⁸⁴PPEY²⁸⁷ motif of Amot-p130 through its N-terminus [23]. Whether KIBRA mediates Motin function is not known, yet this protein is involved not only in regulation of the Hpo-YAP pathway [80-82] but also in the regulation of epithelial cell polarity [83-85]. Future analyses will help to clarify the significance of this interaction.

A number of recent studies have described the negative regulation of YAP by members of the Motin family [33, 34, 37, 47, 48, 72] (Figure 4). Regarding YAP/TAZ subcellular localization, it has been suggested that upon ectopic expression of Angiomotins in HeLa and MCF7 cells, YAP and TAZ translocate from the nucleus to the cytoplasm and in MCF10A cells, Amot-p130 co-localized with the actin cytoskeleton [37, 47, 48]. Interestingly, the expression of exogenous Angiomotins in some of these systems resulted in the localization to cytoplasmic puncta instead of targeting to TJs [47]. Furthermore, in the tight junction-forming MDCK cells, Amotp130 expression induced partial translocation of YAP to the cytoplasm and co-localization with the TJ protein ZO-1 [47]. Loss-of-function analysis in cell lines expressing at least one of the three members of the family added further insights into Motins involvement in the negative regulation of YAP function. Consistent with findings from overexpression studies in MDCK or MCF10A cells, the knockdown of either Amot-p130 or AmotL2 not only caused loss of TJs with augmented localization of YAP and TAZ in the nucleus, but also promotion of Epithelial-Mesenchymal Transition (EMT) and up-regulation of YAP downstream target genes CTGF and Cyr61 [34, 37, 47, 48]. Noteworthy, attenuation of AmotL1 expression was refractory to these phenotypic and transcriptional changes, pointing out the distinct behaviors between Motin family members. Interestingly, in H441 human lung cancer cells, AmotL2 induced a robust shuttling of TAZ from the nucleus to the cytoplasm [72]. This highlights cell-context dependency, suggesting the tissue or organs wherein they are expressed determines Motin functions (Figure 2).

Fig. 4. Proposed models for YAP regulation by the Motins.

Amot-p130 as a positive regulator of YAP: (1) Amot-p130 interaction with YAP prevents its association with LATS1/2 in the cytoplasm and LATS-mediated YAP phosphorylation. Subsequently Amot-p130:YAP complex translocates to the nucleus to promote the transcription of TEAD-target genes. Amot as a negative regulator of YAP: (2) LATS-mediated phosphorylation of Amot promotes Amot-YAP association and subsequent inhibition of YAP activity. Whether Amot phosphorylation regulates YAP sequestration to the cytoplasm, and/or localization at the TJs is currently unknown. (3) Amot acts as a scaffold protein at the TJ, bringing together MST and LATS, promoting LATS activation and subsequent LATS-mediated YAP phosphorylation. (4) Amot physically interacts with YAP leading to localization at the TJ or to the actin cytoskeleton, independent of LATS-mediated phosphorylation events.

Models for the negative regulation of YAP/TAZ activity by the Angiomotins suggest a number of possibilities leading to localization of YAP/TAZ to the cytoplasm/cell junctions. As discussed previously, a major mechanism of YAP/TAZ regulation is through exclusion from the nucleus. This is regulated to a significant extent through YAP and TAZ phosphorylation sites Ser127 and Ser89 residues, respectively, leading to cytoplasmic sequestration through 14-3-3 binding [86, 87]. Both YAP and TAZ are kinase substrates of LATS1/2 [76, 86]. Once activated, LATS1/2 phosphorylates YAP on five HxRxxS consensus sites – Ser61, Ser109, S127, S164, and Ser381 [88], and TAZ on four HxRxxS motifs – Ser66, Ser89, S117, and Ser311 [89]. Phosphorylation of YAP Ser127 and Ser381 residues by LATS1/2 are key events for its inhibition [88] whilst on TAZ, phosphorylation of Ser89 and Ser311 sites were shown to inhibit its transcriptional activity [89].

Accordingly, it has been suggested that Motins serve as a scaffold for Hippo pathway kinases and YAP at TJs, leading to activation of LATS2 and phosphorylation of YAP2 (Figure 4) [34]. Yet other studies show only mild up-regulation of phosphorylated YAP in HEK293 cells upon exogenous expression of AmotL2 [47]. Other models suggest that Motin-mediated localization of YAP/TAZ to the cytoplasm/cell junctions is independent of YAP/TAZ phosphorylation by LATS1/2 [47]. This is supported by studies showing that this localization it is still observed with YAP^S127A and TAZ^S89A mutants (Figure 4) [37, 48].

In contrast to the findings ascribing a YAP-inhibitory role for Angiomotins, we have recently found that Amot-p130 is required for YAP function both in vivo and in vitro. By using both gain- and loss-of-function approaches, we find that in multiple cell types Amot-p130 directly interacts with YAP and functions as a positive regulator by antagonizing YAP interaction with LATS1, promoting YAP dephosphorylation and translocation to the nucleus (Figure 4) [7]. The differences between the observations described above might reflect use of different cell lines or experimental setups. For example, a potential caveat is the expression of exogenous Motin family proteins tagged at their C-terminal PDZ binding domain, which might mask interactions with binding partners such as Patj/Mupp1. Yet another possibility is that previously unappreciated post-translational modifications play a role in regulating the function of Motin family members. Interestingly, recent studies suggest that Motin-YAP association might also be regulated by phosphorylation of the Motins by LATS1/2 (Figure 4). These studies show that phosphorylation regulates Motin subcellular localization and activities related to cell proliferation and migration [23, 30, 32, 33]. Whether this phosphorylation induces translocation of YAP/TAZ from the nucleus to the cytoplasm and subsequent localization to TJs remains to be determined.

Finally, in addition to mechanisms involving the sequestration of Yap out of the nucleus, we have recently identified a nuclear function for Amot-p130 in regulating YAP activity [7]. Amot-p130 was found in the nucleus where it forms a complex with YAP and the DNA-binding transcriptional factor TEAD1. Gene expression profiling further supported these findings and demonstrates a strong concordance between Amot-p130 and YAP transcriptional profiles, suggesting functional cooperation. In fact, microarray analysis and gene set enrichment analysis (GSEA) indicated that 99.5% of the genes co-regulated by Amot-p130 and YAP are regulated in the same direction, with significant correlation between the top-ranked Amot-p130 and YAP-regulated genes. Luciferase assays and Chromatin immunoprecipitation (ChIP) coupled to RT-PCR show that in HEK293 cells, knockdown of Amot-p130 expression inhibited the binding of a YAP^S112A mutant to the promoter of ApoE (which encodes Apolipoprotein E). Surprisingly, binding of YAP^S112A to the CTGF promoter was not affected by Amot-p130 knockdown. Sequential ChIP analysis further corroborated that Amot-p130 functions as a co-factor in a transcriptionally active YAP-TEAD1 complex, involved in recruiting the complex to promoters of a subset of YAP-regulated genes, which does not include CTGF [7, 90].

Function: Angiomotin involvement in the pathological setting

The relationship of Angiomotins to the pathogenesis and progression of human diseases is still poorly understood. The majority of investigations have centered upon the expression of Amot in a variety of cancers and its role in modulating angiogenesis, an essential process for cancer progression and survival. A mouse model of hemangioendothelioma that overexpresses Amot-p80 resulted in fast-growing invasive tumors that were larger than controls [91]. In contrast, an Amot mutant-expressing cell line in which the C-terminus PDZ-binding motif was deleted produced smaller, dormant tumors that displayed a high level of apoptosis. The Amot mutant tumors remained dormant for more than 21 days and did not invade surrounding tissues. Both Amot and deletion mutant tumors were vascularized, but the deletion of the C-terminal amino acids resulted in inhibited cell migration, suggested as an explanation for the lack of invasiveness [91].

Subsequent screening of human breast tumors has revealed significantly increased expression of Amot in tumors relative to normal tissue [92]. In contrast, AmotL1 and AmotL2 have shown no significant correlation with tumor burden. Furthermore, Amot overexpression correlated with tumor grade, decreased survival, and metastatic disease. Taken together, these findings imply an important role for Amot in the context of tumor progression and underscore the suitability of Amot as a potential biomarker and a target for anticancer therapeutics.

Efforts thus far to design Amot-based therapies have centered on immunological-based strategies. A DNA-based vaccine against human Amot-p80 produced in mice has shown promising results in a mouse model of breast carcinoma [18]. Implantation of mouse TUBO (Turin-Bologna) mammary carcinoma cells produced rapidly growing tumors in mice vaccinated with empty vector. By contrast, vaccination with Amot-p80 inhibited tumor growth completely in the majority of mice. In the few examples where tumors did develop the tumors grew much slower compared to control mice. The significant suppression of TUBO tumor growth in response to Amot-p80 vaccination was partially abrogated in CD4-depleted mice and completely nullified in B cell knockout mice, suggesting vaccination effects are antibody-dependent. Amot-p80 vaccination was also tested using a more aggressive transgenic breast cancer model, based on Her-2/neu oncogene under the control of the mouse mammary tumor virus promoter (MMTV). Vaccination with Amot-p80 slightly delayed tumor progression, but when combined with vaccination against Her-2 extracellular and transmembrane domains, a synergistic effect was observed and 80% of vaccinated mice survived for more than 70 weeks, compared to the complete lethality at 20 and 25 weeks for control and Amot-alone vaccination, respectively [18, 93]. Significantly, the expression of Amot was observed to increase with the progression of neoplastic lesions to invasive carcinomas [93]. Vaccination with Amot-p80 inhibited angiogenesis in implanted tumors, and isolated Amot antibodies inhibited the migration of mouse aortic endothelial cells, replicating the effects of angiostatin. Finally, Amot-p80 vaccination resulted in increased permeability of TUBO tumors, which increased the efficacy of chemotherapy [93].

Interestingly, Amot expression is found to be significantly upregulated in schwannomas from NF2 patients, which displayed a primarily localization of Amot to the nucleus [7]. Further analysis of a role for Amot in this tumor type found that Amot is essential for the development of schwannomas by Nf2-null Schwann cells [43]. Using an orthotopic model of NF2-associated schwannoma, luciferase-expressing Nf2-null schwann cells (SC4) that stably express Amot shRNA were implanted intraneurally into the sciatic nerve and tumor progression was monitored via bioluminescent imaging. In contrast to control SC4 cells, which formed substantial tumors one week after implantation, Amot knockdown resulted in a significant impairment in the growth of schwannomas. Western blotting revealed sustained suppression of Amot expression and MAPK signaling, which explains the slower tumor growth and is consistent with the proposed mechanism. Together, these results demonstrate the requirement of Amot for development of schwannoma in response to loss of Nf2.

Further highlighting the interplay between Nf2, Amot and the Hpo-YAP pathway in pathological scenarios in vivo is the finding that Amot-p130 enhanced the activity of Yap to promote tissue repair and liver tumorigenesis [7]. Amot conditional knockout (KO) mice were crossed with albumin-Cre transgenic mice generating a mixed AmotKO (Alb-cre:Amot^flox/flox) genetic background with liver-specific deletion of Amot. After 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DCC)-induced injury, it was observed that in response to injury the proliferation of biliary epithelial cells (BEC) was impaired in the Alb-cre:Amot^flox/flox mice. The analysis was extended to liver-specific Nf2 knockout mice and Nf2:Amot double knockout (DKO) mice. These studies established a requirement for Amot in hepatic ductal cell proliferation and tumor formation in the context of Nf2 loss. In addition, a significantly increased expression of Amot was observed in Nf2-null liver tumors [7].

Interest in Amot's role in angiogenic-related pathology has revealed several new disease interactions. Angiogenesis is known to play a pivotal role in age-related macular degeneration, with VEGF driving neovascularization of the choroid, leading to retina damage [17, 94, 95]. VEGF inhibitors are approved for use in macular degeneration, but with compensatory angiogenic pathways allowing resistance [96, 97], there is significant interest in targeting other drivers of angiogenesis. In a mouse model of macular degeneration, intraperitoneal injections of PEGylated Amot Fab reduced choroidal plaques by 73% [17]. Finally, recent research has correlated Amot expression to microRNA-induced tumor dormancy [98]. In this disease state, proliferation is balanced by increased apoptosis and an impairment of angiogenesis. Dormant glioblastoma multiforme tumors that expressed dormancy-associated miRs also showed elevated levels of Amot with enriched angiostatin, suggesting a potential antiangiogenic mechanism for tumor dormancy.

Overall, the identified interactions of Amot within disease pathways present a complex picture, with angiogenic signaling and Hpo-YAP pathway interactions both driving the pathology of a variety of tumor types. It is clear from the genetic and immunologic studies that Amot is a promising therapeutic target and future work will undoubtedly reveal a more thorough and cohesive picture of Amot dysregulation and interactions within a wide range of tumor and angiogenic disease types.

Open questions

There are major gaps in our understanding of the Motin family of proteins as detailed throughout this review. One of the major open questions relates to functional redundancy between members of the family. While there are clear areas of overlap it is also clear that there are non-redundant roles that are likely to be cell-type specific. Another critical question is which are the relevant pathways through which the Motins exert their functions. The discovery that the Motins orchestrate the function of multiple signaling pathways, including pathways regulated by small G-proteins and the Hpo/YAP pathway, draws a complex picture. Interestingly, the majority of studies examining Amot in animal tumor models and human tumor samples suggest Amot expression is upregulated and, in some cases, required for disease progression. This raises questions regarding which Amot-related pathways are involved and how these pathways modulate Amot functions within the context of the disease. Again, this is also likely cell-type and context-dependent. Further complexity has been recently introduced with the identification of post-translational modifications that likely regulate the function of the Motin themselves. The role of these modifications, as well as other transcriptional and post-transcriptional mechanisms, are only beginning to emerge and ongoing research by multiple groups will surely lead to a better understanding of these mechanisms in short order.

Given the underlying complexity of Motin function, these questions will be most accurately addressed by employing a combination of molecular and genetic approaches, in relevant cell types and whole organism models.